586 
WALTORE 
. 
tinacity, it is not so integral a part of its spectrum as the other 
red, green, and orange lines. For instance, the chemical as well as 
physical behaviour of these line-forming bodies is different. On 
closely comparing the spectra of specimens of samaria from dif- 
ferent sources, line S6 varies much in iatensity, in some cases 
being strong and in others almost absent ; the addition of yttria 
is found greatly to deaden the red, orange, and green lines of 
samarium, while yttria has little or no effect on the line Sé; 
again, a little lime entirely suppresses line S8, while it brings 
out the samarium lines with increased vigour. Finally, attempts 
to separate line Sé from samarium and those portions of the 
samarskite earths in which it chiefly concentrates has resulted in 
sufficient success to show me that, given time enough, and an 
almost inexhaustible supply of material, a separation would not 
be difficult. These facts, together with the peculiar behaviour 
of the lines Ge and Gy, strengthen my suspicion as to the 
resolvability of samarium. 
Samaria giving the line S5 had been prepared from cerite and 
samarskite. Many observations had led me to think that the 
proportion of band-forming constituents varied slightly in the 
same earth from different minerals. Amongst others, gadolinite 
showed indications of such a differentiation, and therefore I con- 
tinued the work on this mineral. Very few fractionations were 
necessary to show that the body giving line S85 was not pre- 
sent in the gadolinite earths ; no admixture of yttria and samaria 
from this source giving a trace of it. It follows, therefore, that 
the body whose phosphore:cent spectrum gives line $5 occurs in 
samarskite and cerite, but not in gadolinite. 
It now became an interesting inquiry whether all these 
constituents of yttrium were united together in exactly the 
same proportion in every case. A glance at the diagram will 
show that yttrias from different sources, although they may be 
alike as far as our coarser chemical tests are concerned, are 
not built up exactly in the same manner. Thus, when the 
samarskite yttrium was forming, all the constituent molecules— 
which I have provisionally named Ga, GB, Gy, G3, Ge, G6, 
Gy, and Sé—condensed together in fair proportion. In gado- 
linite yttrium the constituents G8 and Gé are plentiful, G¢ is 
very deficient, S6 is absent, and the others occur in moderate 
quantities. In the yttrium from xenotime Gé is most plentiful, 
GB occurs in smaller proportion, Gis all but absent, and S5 is 
quite absent. Yttrium from monazite contains GB and G65; with 
a fair proportion of the other constituents, G8 is plentiful and 
the red is good. Yttrium from fluocerite is very similar to that 
from monazite, but Ga is weaker. Yttrium from hielmite is 
very rich in Gd, has a fair quantity of Ga and Gf, less of Gy, 
no Sd, and only a very faint trace of Gn. Yttrium from euxenite 
is almost identical with that from hielmite. Yttrium from cerite 
contains most G¢ and G8, less Ga and G8, only a trace of Gn, 
anda fair proportion of S5. 
I have already mentioned how the key to these explanations 
was gained by an examination of the phosphorescent spectrum 
of M. de Marignac’s Ya (now called by him gadolinium). 
Referring to the diagram, it is seen that Ya is composed of 
the following band-forming bodies :—GB£, $3, G¢, together with 
alittle samarium. Calling the samarium an impurity, it is thus 
seen that gadolinium is composed of at least three simpler bodies. 
It is by a method of his own, differing from mine, that M. de 
Boisbaudran has obtained phosphorescent spectra of some of the 
rare earths. He takes the induction-spark between the surface 
of a strong and acid solution of the metallic chloride and a clean 
platinum wire a few millims. above it. The platinum wire is 
kept negative and the solution positive ; it is then ob-erved that 
in many cases a thin layer of fluorescent light is seen at the 
surface of the liquid. This layer gives a spectrum of nebulous 
bands. For the sake of brevity I will adopt M. de Boisbaudran’s 
term, and call this process the method of reversion (the direction 
of the spark being reversed). As this method is entirely different 
to the one I adopt, it is not surprising that the results are also 
different. Experimenting in this way M. de Boishaudran has 
obtained, among others, two bands (A 573 and A 543°2), which 
he considers are caused by two elements, named respectively Za 
and ZB, and which he considers new, at all events if we except 
terbium and pos-ibly the elements of what was formerly called 
holmium. His method fails to show any spectrum in solutions 
of yttria which by my method give the yttria bands with the 
greatest brilliancy ; while conversely his method shows a fluor- 
escent spectrum in solutions of earths separated as widely as 
possible from yttria, chemically as well as spectroscopically. My 
experiments on both these methods tend to the conclusion that 
| 
our bands are not due to the same cause, although M. de Bois- 
baudran’s experiments have led him to the opposite conclusion. 
The band of ZB (543) falls between the double green band G8, 
and the band of Za (573) would come very near the citron line 
G6. 
In the hands of a practised experimentalist like M. de Bois- 
baudran this method may give trustworthy indications, but I 
must confess that in my opinion the test is one beyond the range 
[Oct. 14, 1886 
of practical analysis, owing to the enormous difficulty of getting — 
the phenomena described by the discoverer. Unless the strength 
of spark, the concentration and acidity of solution, and the dis- 
persion and magnifying power of the spectroscope bear a certain 
ratio one to the other, the observer is likely to fail in seeing a 
spectrum even in solutions of earths which contain considerable 
quantities of Za and Z8. In my own case I not only have had 
the advantage of personal instruction in Paris from M. de Bois- 
baudran himself in the best method of getting these reversion 
spectra, but on returning to London I brought with me some of - 
the identical earths which give these spectra at their best. 
spite of these advantages I have sometimes experimented off and 
on for weeks without being able to see more than a feeble 
glimmer of the bands described by M. de Boisbaudran. 
Again, when everything is most favourable and the reversion 
bands are at their strongest, they are but a faint and hazy 
shadow of the brilliant lines given by the bombardment process. 
M. de Boisbaudran, speaking of the relative sensitiveness of our 
two methods, says that the bombardment process 7” vacuo is 
incomparably more delicate than his reversion test, and I esti- 
Ing 
oo 
mate the relative sensitiveness of the two methods to be in the © 
proportion of about I to Ioo. 
You have probably anticipated in your minds a question which 
is likely to occur at this point of the inquiry. If such results 
have been obtained by submitting yttrium to this novel method 
of analysis, what will be the result of fractionating some other 
reputed element ? 
Yttrium, as I have explained, is an exczedingly stable mole- 
cular group, capable of acting as an element, just as calcium, — 
for instance, acts as an element : to split up yttrium requires not 
only enormous time and material, but the existence of a test by 
means of which the constituents of yttrium are capable of recoz- 
nition. Had we tests as delicate for the constituent molecular 
groups of calcium, this also might be resolved into simpler 
groupings. It is one thing, however, to find out means of 
separating bodies which we know to be distinct and have colour 
or spectrum reactions to guide us at every step; it is quite 
another thing to separate colourless bodies which are almost 
identical both in chemical reaction and atomic weight, especially 
if we have no suspicion that the body we are dealing with is a 
mixture. 
(I mention calcium because it is one of several other elements 
which I have put through the fractionation miil. Many hundred 
operations have given me just sufficient encouragement to make 
me wish I had time to push this work to the end.) 
One of the chief difficulties in the successful carrying out of an 
investigation in radiant-matter spectroscopy is the extraordinary 
delicacy of the test. This extreme sen<itiveness is a drawback 
rather than a help. To the inexperienced eye 1 part of 
yttrium in 10,000 gives as good an indication as I part in [0, 
and by far the greater part of the chemical work undertaken 
in my hunt for spectrum-forming elements was performed 
upon material which later knowledge shows did not contain 
sufficient to respond to any known chemical test. It is as if the 
element sodium were to occur in ponderable quantity only in a 
few rare minerals seldom seen out of the collector’s cabinet. 
With only the yellow line to guide, and seeing the brilliancy 
with which an imponderable trace of sodium in a mineral de- 
clares its presence in the spectrum, I venture to think that a 
chemist would have about as stiff a hunt before he caught his 
yellow line as I have had to bring my orange and citron bands 
to earth. ; 
Chemistry, except in few instances, as wate--analysis and the 
detection of poisons, where necessity has stimulated minute 
research, takes little account of ‘‘ traces,” and when an analysis 
adds up to 99°999, the odd o‘oor per cent. is conveniently put 
down to “impurities,” ‘‘ loss,” or ‘* errors of analysis.” When, 
however, the 99°999 per cent. constitutes the impurity, and this 
exiguous ‘oor is the precious material to be extracted, and 
when, moreover, its chemistry is absolutely unknown, the diffi- 
culties of the problem become enormously enhanced. Insolubility 
as ordinarily understood is a fiction, and separation by preci 
diel nate tie 
